Re: Building my own equilizer

[Mike Thompson <mikt@pair.com>]

> At higher currents, battery internal resistance, wiring resistance, and

Battery internal resistance is a function of battery state of charge, so
you'll see some additional voltage differential due to this under
discharge. Though it's in the wrong direction during charge. Connection
resistance, not a function of SOC, will dominate in some cases.

> connector resistances cause voltage drops that are not related to state
> of charge. Under load, you can easily have a battery at a higher state
> of charge with a lower terminal voltage due to these resistances.
>
> Suppose your batteries have 5 milliohms +/- 1 milliohm of resistance.
> Driving at 100 amps, that 1 milliohm variation causes 0.1v of variation
> in voltage drops. This means the voltage difference due to variations in
> states of charge get lost in the noise. A balancer that depends solely
> on voltage can charge the wrong battery. until the difference is quite
> large.

A problem likely to affect most anything that equalizes during charge or
discharge using voltage as an indicator. A tiny PIC controller with a
little memory of history would address that in our dream battery
management system of the future...

> > I haven't seen any data or theoretical argument that the PowerCheqs
> > are incapable of bringing the batteries into balance within a week
> > or two of installation. Driver experience indicates that they do
> > keep the voltages balanced.
>
> I have seen the Powercheqs, read their data, and talked to them at trade
> shows. As far as I know, they work as advertised. I don't think there is
> anything wrong with them per se.
>
> Likewise, I have seen and even tested the Zivan Smoother, and have
> literature on the Badicheq. There is nothing seriously wrong with any of
> them. They all push your batteries toward balance. This is good. If you
> measure performance by "how close battery voltages match", then they all
> work.
>
> But here's my fundamental point:
>
> 	BATTERY VOLTAGE IS A POOR INDICATOR OF STATE OF CHARGE

Well, of course. unfortunately battery management is in a very primitive
state, and until we have some better answers at suitable price points,
many systems will be using battery voltage in one way or another for
better and worse.

> This is something that anybody who's tried to use voltage for a state of
> charge gauge knows. All these balancers (and mine too) use voltage as if
> it is the same thing as state of charge. But it's not. It's a fairly
> good indicator when 0 current has been flowing for a long time, but gets
> progressively worse as current in/out increases or changes quickly.
>
> So, voltage based balancers tend to make bad decisions while driving or
> charging. They are likely to charge the wrong battery.

They also are likely to extend your range a few percent when you need
it most. If they had some history (tiny pic as above) to keep them from
making the wrong decision and get a rough approximation of how much
current to move, that would be very powerful compared to the alternatives.

> Luckily, EVs sit idle for long parts of each day. During these times,
> the balancers *will* work to balancer the voltages.
>
> > You're assuming you know the PowerCheq current transfer as a function
> > of the battery module voltage difference.
>
> Yes, I am making assumptions. But they are based on what I know of their
> circuit, and on the observed current and voltage seen in their demos.
>
> > Since this is a high frequency switching design of some
> > sophistication, they could design it to whatever current they want
> > at a 50 mV module voltage difference. Not like a typical linear
> > circuit or resistor where the current goes to zero as the voltage
> > difference approaches zero.
>
> Remember that the BATTERY is what determines the current for a given
> voltage. And I know a lot about batteries :-)  To raise the voltage of a
> 12v battery at 0 amps by 50mv, I guarantee the current will be an
> exponential decay, starting very briefly at whatever the power source
> will deliver, but rapidly falling to less than 1% of the battery's
> amphour capacity.

Hmmm.... Better add some rudimentary brains to the balancers with a few
bits of history.

Not that many good choices for battery management yet...

> >> I said it was a simplistic, worst-case argument. But I could equally
> >> well have assumed 200ah batteries being 50% discharged. It would have
> >> complicated the math without providing any additional insight.
>
> > So? I guess your EV has 200 Ah batteries?
>
> My last EV had 6v 220ah golf cart batteries. My current one has 12v 95ah
> batteries, which is why I used 100ah for my example.
>

Ahh... so you have! You could double the number of PowerCheqs or over
balancers... at more cost of course :-(

> >> You start out with both at full charge and 13.0v no-load. You apply a
> >> 40-amp load for 1 hour, so you've removed 40ah...
>
> > Without PowerCheqs during load. Let me presume that for 30 minute of
> > that time, the voltage difference is 120+mV so the PowerCheqs are
> > running at full 2A equalization for 30 minutes during the discharge.
> > That means 1/2 hour * 2A = 1 Ah is transferred from the strong battery
> > to the weak battery WHILE the discharge is happening!
> >
> > battery#1: 110ah - 40ah - 1ah = 69ah
> > battery#2:  90ah - 40ah + 1ah = 51ah
>
> Yes, they could provide another 1ah. But it's not much compared to the
> 40ah. And as mentioned above, it might be going into the wrong battery.
> So for simplicity, I ignored it.

Not much is in the eye of the beholder! You know how hard we work and pay
for each ah!

so 1ah/40ah = 2.5% . Not earth shattering but worth an extra mile or two.

If it were free, I doubt anyone would turn it down! Of course none of
these charge management systems is free.....

> > the PowerCheqs work quite nicely especially compared to most other
> > equalization methods which just let the batteries continue to drift
> > out of balance until the next charge cycle.
>
> See my "point" at the beginning of this email. What I am trying to do is
> figure out how many amphours of balancing we might actually get with the
> Powercheqs in this example. Then you have a basis to compare it to other
> methods of balancing. For example, a Zivan charger that equalizes on
> every cycle, putting in a couple extra amphours no matter what to
> accomplish balancing. Or Rudman regulators, that do all their balancing
> in the last hour or so of a charge cycle.

PowerCheqs can also be used with the equalizing cycle of the charger. It
does appear that all of the readily available methods will still need some
lower current equalization charge from the charger at the end of the
cycle. Though as the pack comes into balance, it seems this period can be
pretty short 15-60? min as opposed to 2-4 hours for a pack without active
battery management.

> > battery#1: 66ah - 40ah - 1ah = 25ah; 25/110 = 23% SOC; 12.23v resting
> > battery#2: 54ah - 40ah + 1ah = 15ah; 15/90  = 17% SOC; 12.17v resting
>
> >> The charger, which knows nothing about any imbalance, will see only
> >> the total voltage; 12.24v + 12.16v = 24.4v. It will assume the state
> >> of charge is 12.2v or 20%. So it puts in 80ah.
>
> > Puhlllleeeze Lee! Do you REALLY own a charger that guesses how many ah
> > were removed based on the series string voltage???? ...
>
> Calm down, Mike. (You really should cut back on your caffeine)!

WHERE'S MY COFFEEE!!!

> All I'm saying is that we don't know anything about the charger's
> algorithm. It is not relevant to the discussion. But we DO know that it
> is going to put some number of amphours in, the same amount to every
> battery. And, for essentially all chargers, the number of amphours
> varies depending on the battery's state of charge (if it's half charged,
> it will put about half as many amphours in). So I'm just saying that by
> whatever means, the charger put 40ah in. Again for simplicity, I assumed
> 100% efficiency.

Where I disagree is that the charger "assumes" the state of charge. It's
not that smart, we both know that! Most chargers stuff ah in until the
pack hits a certain voltage and then back off the current. Most don't
remember the voltage the charge started at and use that to calculate ah.

Without a smarter battery balancer using voltage and history, it's
becoming clear to me that the chargers taper / leveling phase is still
required for virtually any method.

> The important point is that the batteries are not balanced at this time.
> The balancer has not had time to correct for the imbalance caused by
> driving.
>
> >> The Powercheq won't help during charging, for two reasons. First,
> >> because its current is small compared to the charging current.
>
> > Yeah right. An EV charger is often 10-30 amps for 100V packs and up.
> > The PowerCheq current (max) is 2A or 20%-7% of that.
>
> I guess it depends on whether you consider 7% to 20% "small."

Compared to the imbalance and typical charge times, it's significant.

> >> second, it will put charge into battery#2 early in the charge cycle
> >> (since it is initially the lowest voltage), but then put charge into
> >> battery#1 later in the cycle, when battery#1 voltage becomes lower.
>
> > True. It will mistakenly put charge current into the weaker battery,
> > but that will be less than half of the bulk charge phase...
>
> Right. All I'm showing is that voltage-based balancers make mistakes
> during high-current bulk charging. But they correct them if given enough
> time after the charger tapers down or shuts off.

Must have history!!
Something that guesses battery capacities at rest or at some other
opportune time and then remembers that. Imagine the balancers just having
a list of batteries from weakest to strongest and some low resolution
indicator of their capacity.. Even 4 bits (16 = 100%, 0=80%, below 80%
you've got big problems anyway and the battery will still get special
treatment...) The balancers would also do well knowing pack current, so
they can combine rough Charge / Discharge current, history with rough
relative capacity estimate (good to bad in 16 steps as above). NOw you've
got a balancer that will always take from the high capacity and give to
the low capacity and even do pretty well and giving the most help to the
batteries that need it most.

Take the powercheqs, add this info to them, connect them with a bus, have
them measure voltage, a small central controller controls them and knows
pack current, double the powercheq balance current to 4A (for up to 100ah?
200ah? batteries) and with their new found intelligence, they could drive
4A from the strong battery to the weak battery even when that current
causes the weak battery voltage to exceed the strong battery because the
balancers are smart now and learn from history!

> > The only thing that eliminates this level of imbalance for a charge
> > cycle without a couple hours of equalization is your [Battery
> > Balancer] or shunt regulators like Rudman's with external load
> > resistors. But wait, the MK3's actually throttle the charger back.
> > So most things that deal with such severe imbalances have the
> > charger do some taper / equalization charge at the end!
>
> Right! If there's a large imbalance, you can:
>
> a) Do an equalization cycle (deliberately overcharge the good ones to
>    bring up the low ones -- hard on sealed batteries)
>
> b) Use a balancer with a "strong" per-battery correction current (like
>    my Balancer, or Rudman regulators with big resistors).

Strong per battery current could even do this during the bulk charge if
voltage confusion is fixed with history. Though if the pack has a
distribution of capacities the strong current divided by the attention of
several weak batteries can effectively reduce the strong current of a 1
battery at a time equalizer down to that of the many at a time equalizer
with a small current. Dependent on the positions of the weak / strong
batteries in the string...

> c) Use a balancer with a "weak" per-battery correction current (like
>    the Badicheq, Powercheq, or Smoother), and throttle back the charger
>    so it doesn't equalize the batteries before the balancer has had
>    time to correct the problem.
>
> >> Yes; that's what happened to me on one occasion. The Balancer hid the
> >> fact that I had a bad battery. The range is limited by the average
> >> capacity of the batteries in the pack rather than by the weakest one.
>
> > As shown above, please include that in future analysis of always on,
> > charge shuttling battery equalization systems. I can't fathom why you
> > left that out if you're really thinking about the problem.
>
> A balancer only hides a weak battery if it is strong enough to
> compensate for that battery's loss in amphour capacity. If the battery
> is 10ah less, then the balancer needs to put 10ah into it during a
> 1-hour drive to keep it from going dead early and limiting your range.

But you're likely to already be aware of your limited range. The active
during discharge balancers improve that situation over a solution that is
inactive during discharge. I'd take the benefit every time if it didn't
cost much.

> > I guess if you like driving with a pack that's 20% out of balance,
> > your equalizer could give you an extra 10-15 Ah during driving.
>
> That's what I'm trying to do. I'm a cheapskate. I want to use my
> batteries as long as possible, until they won't do the job any more. A
> "strong" balancer has the potential to extend battery life, by extending
> range and letting me keep driving on batteries that otherwise would be
> too imbalanced to use.

Ah, so your range is lower than it would other wise be... But that's just
fine if it meets your needs since we all have different requirements.

> > Why don't you just tell me straight out why you despise PowerCheqs?
>
> I don't "despise" the Powercheqs! They just seem expensive for what they
> do, and a bit naive in their claims.

Especially since you are capable of building your own!

The PowerCheqs do need a serious cost reduction and some additional
intelligence. Unfortunately there aren't that many alternatives expensive
or cheap. Rich is probably making the most progress in terms of increasing
intelligence and keeping cost in reason. The PowerCheqs have been
relatively stagnant.

>
> > But your battery balancer is unobtanium unless someone contacts you.
>
> Am I that hard to contact? My address is in every email. I've sent
> dozens of sets of plans out for $5 to anyone who wants them. That's
> cheap! I don't think I could be any more open in sharing the design.
>
> I designed my Battery Balancer for my own use, as a research tool to see
> if the idea works. Because I only intended to build a few, it uses parts
> and construction techniques best suited to build one at a time. Thus it
> would be a very costly design to put into production.
>
> I don't care, and even encourage others to duplicate it. In fact, I've
> sold my extra parts kits and PC boards at cost. My hope was that others
> would share in the development. So far, that hasn't happened. Selling
> them just means I get more calls and emails requesting (free) help.
>
> > Maybe you can share your ideal solution and we can all start an
> > obtainable implementation of it? We don't have that many available
> > low cost methods of battery equalization (barring equalization
> > charges of floodeds) available to us.
>
> I *am* sharing what I've learned! (And getting lots of complaints for
> not giving away perfect solutions for free :-)
>
> > I'll happily add the best solutions to my battery management page. The
> > PowerCheqs and Rudman MK3's are already there. I tried to put your
> > solution there but it was unobtanium when I tried to get it. Maybe
> > that's changed?
>
> I have no objection to you or anybody else posting my Battery Balancer
> schematics and parts lists. People just need to be aware that it is a
> research tool, not a finished product. They have to build it themselves,
> write their own software, adapt it to their own battery pack, decide on
> their own algorithms for using it, and debug it all. I can't provide
> infinite free advice!

Cool. Want to send it to me? I'll put it on my website. I'll put your info
however you want - none, Lee Hart, email only, whatever you like. We all
hear so much about your balancer it's frustrating not to know more about
it.

Last time I asked we didn't quite connect. Let me know how you want it
posted and I'll do it that way.

> --
> Lee A. Hart                Ring the bells that still can ring
> 814 8th Ave. N.            Forget your perfect offering
> Sartell, MN 56377 USA      There is a crack in everything
> leeahart_at_earthlink.net  That's how the light gets in - Leonard Cohen
>
>

Later,          "A rainbow is only part of a circle." San Joser, CA
     ^ ^        Software Consultants:        http://www.migration.com/
     O o        Permanent:                   m.t.thompson@ieee.org
  ===-o-===     My catbox:                   http://www.madkatz.com/
Ack! Phththpph!

2001 Clean Air Champion http://www.baaqmd.gov/pie/press/cachamp01.pdf

My True Zero Emission Vehicle (TZEV) GM EV1 and my home are powered
by the 100% renewable, Zero Emissions Electricity (ZEE) 30kWh/day
Solar Electric (PV) system on my roof:
http://www.madkatz.com/pv/index.html

Sometimes I see gas cars... In my rearview mirror! http://www.gmev.com/

President, Electric Auto Association, San Jose Chapter
http://www.geocities.com/sjeaa/

EV List Archive Owner,Admin  http://www.crest.org/discussion/ev/current/
Electric Vehicle Email List Instructions: (EVList members: Bookmark it!)
http://www.madkatz.com/ev/evlist.html